Missing dependent variables in panel data models

Size: px
Start display at page:

Download "Missing dependent variables in panel data models"

Transcription

1 Missing dependent variables in panel data models Jason Abrevaya Abstract This paper considers estimation of a fixed-effects model in which the dependent variable may be missing. For cross-sectional units with some but not all) dependent variables missing, covariate information from all time periods can be utilized to improve efficiency of estimators. Estimation of fixed-effects models with exchangeability where the fixed effect is independent of the ordering of observations for a cross-sectional unit) and lagged dependent variables are also considered. Keywords: missing data; fixed effects; linear projections. Financial support from the National Science Foundation SES ) is gratefully acknowledged. This paper has benefited from comments by Stephen Donald and seminar participants at Northwestern, UBC, and the Midwest Econometrics Group meeting. Department of Economics, The University of Texas at Austin; abrevaya@eco.utexas.edu.

2 1 Introduction Data missingness is a common problem in empirical research. In this paper, we focus on the case in which data on the dependent variable may be missing. In a cross-sectional setting, unless one is willing to adopt an imputation approach, the standard approach is to drop observations for which the dependent variable is missing. Such an approach does not affect consistency of the resulting estimators if the model s disturbances are unrelated to the missingness mechanism. 1 The approach of dropping observations would also be appropriate in a panel-data setting under a similar assumption on the missingness mechanism), but we show that this approach can result in unnecessary efficiency losses. Specifically, in models with fixed effects heterogeneity that is correlated with the covariates), covariate information from all time periods, including those with missing dependent variables, can be incorporated into estimation to improve efficiency compared to standard fixed-effects estimators. The paper is organized as follows. Section 2 introduces the linear fixed-effects model and the dependent-variable missingness mechanism. The Chamberlain 1982) projection approach is reviewed, and the impact of dependent-variable missingness on several fixedeffects estimators is discussed. the presence of missing dependent variables. Section 3 provides a framework for GMM estimation in The GMM estimators have closed-form IV formulas, and the optimal GMM estimator is obtained as a two-step estimator. Section 4 discusses estimation under the additional assumption of exchangeability, where ordering of the observations within a cross-sectional unit is not related to the distribution of the fixed effect. Such an assumption has been considered by Altonji and Matzakin 2005) and may be applicable, for instance, in cases where the cross-sectional unit is a group like siblings) and the time periods are individual group members. GMM estimation under exchangeability can provide efficiency gains and also can identify the primary coefficient parameters in the presence of a single observed dependent variable for each cross-sectional unit. Section 5 relaxes the strict exogeneity assumption and considers estimation when the fixed-effects 1 In the linear regression model, the formal condition for consistency of OLS is that the error disturbance, conditional on observability of the dependent variable, is not correlated with the covariates. [1]

3 model has lagged dependent variables as explanatory variables. Missingness causes further complications here since missingness affects both the dependent and independent variables of the model. 2 The model Consider the standard linear fixed-effects model y it = x it β 0 + c i + u it i = 1,..., n; t = 1,..., T ), 2.1) where i indexes cross-sectional units or groups) and T is the total number of time periods or group members). The case of a balanced panel T not varying over i) is considered for simplicity, although the ideas developed in this paper could also be applied to unbalanced panels. We consider the case of large-n n ), fixed-t asymptotics. Let k denote the number of covariates in x it so that β 0 is a k-vector). To allow for missingness of dependent variables y it ), an observability indicator is defined as follows: { 1 if yit is observed s it 0 otherwise For missing dependent variables s it = 0), we use the convention that y it = 0. The observed data is {s it, s it y it, x it )} T t=1 for i = 1,..., n. Note that x it is always observed.) Without loss of generality, assume T t=1 s it > 0 for each i that is, y it is observed for at least one time period for each i). 2 To simplify exposition, the following stacked quantities are defined: y i y i1,..., y it ), x i x i1,..., x it ), u i u i1,..., u it ), s i s i1,..., s it ). Each of these stacked quantities are column vectors, with x i having kt elements and y i, u i, and s i having T elements. 2 This assumption is merely for notational convenience. With Assumption 1 below, this assumption says that we have already dropped cross-sectional units with all y it missing from the dataset. These observations would be dropped from any of the proposed estimators below. [2]

4 When missingness of y it is not an issue, the standard strict exogeneity assumption for the fixed-effects model 2.1) is Eu i x i, c i ) = 0. The observability indicator s i is incorporated into the conditioning set to yield the strict exogeneity assumption for the current setup: Assumption 1 Strict exogeneity) Eu i x i, c i, s i ) = 0 Note that Assumption 1 allows observability s i ) of the dependent variable to be related in arbitrary ways with x i and c i but restricts the error disturbances u i ) to be conditionally mean independent of x i, c i, s i ). 3 The Chamberlain 1982) projection approach considers the linear projection of the fixed effect c i upon x i1,..., x it ), c i = ψ 0 + x i1 λ 01 + x i2 λ x it λ 0T + a i, 2.2) where ψ 0 is a scalar, each λ 0t is a k 1 vector, and E[x ita i ] = 0 by construction for each t). Plugging 2.2) into 2.1) yields a model from which β 0, ψ 0, λ 01,..., λ 0T ) can be estimated: y it = x it β 0 + ψ 0 + x i1 λ 01 + x i2 λ x it λ 0T + a i + u it. 2.3) An additional assumption on missingness is required for the Chamberlain approach to be applicable to the case of missing dependent variables: Assumption 2 Ignorability of missingness for the projection error) E[x ita i s i ] = 0 for t = 1,..., T A sufficient condition for Assumption 2 to hold is that the random variable c i x i is independent of s i i.e., the conditional distribution Dc i x i ) is the same as the conditional distribution Dc i x i, s i ). Before discussing estimation in the presence of missingness, we briefly review some simple estimators when dependent variables are fully observed. A particularly simple estimation method is the pooled ordinary least squares OLS) regression i.e., OLS of y it on 3 This assumption is the same made by Wooldridge 2002) in the context of unbalanced panel-data models, where missingness s it = 0) corresponds to the full observation y it, x it ) being missing rather than just y it. [3]

5 x it, 1, x i1,..., x it )). It is known that this estimator is numerically equivalent to the within estimator and also a Mundlak 1978) regression estimator), as stated in the following result: Equivalence result for fixed-effects models with no missing dependent variables: The following estimators of β 0 are numerically equivalent: i) the within estimator OLS of ÿ it on ẍ it ), where ÿ it y it T 1 T s=1 y is and ẍ it x it T 1 T s=1 x is, ii) the Chamberlain regression estimator OLS of y it on x it, 1, x i1,..., x it )), and iii) the Mundlak regression estimator OLS of y it on x it, 1, x i )). While these three linear-regression estimators are equivalent in the case of perfect dependentvariable observability, this equivalence breaks down when dependent variables are possibly missing. The following subsection illustrates this point, focusing on the simple case of two time periods T = 2). 2.1 Linear-regression estimators with missing dependent variables We consider the two time-period case T = 2) case for simplicity, although the basic idea of this subsection extends easily to T > 2. Fully observed cross-sectional units have s i1 = s i2 = 1, whereas partially observed cross-sectional units have s i1 = 0 or s i2 = 0. For each of the three linear-regression estimators described above within, Chamberlain, Mundlak), consider the logical extensions to the missing-y setting and the resulting properties of the estimators under Assumption 1). Within estimator: Any partially observed cross-sectional unit i would be dropped since ÿ it is unknown. 4 The resulting within estimator, regressing ÿ it on ẍ it for the subsample of fully observed cross-sectional units, is consistent under Assumption 1. Assumption 2 is not needed since the projection error is eliminated by the within transformation.) 4 For T > 2, within transformations can be used on the subset of s it = 1 observations when t s it 2. [4]

6 Chamberlain estimator: When s it = 0 observations are dropped, the resulting pooled OLS estimator y it on x it, 1, x i1, x i2 ) for s it = 1 observations) is consistent under Assumptions 1 and 2. This modified Chamberlain estimator is not equivalent to the modified within estimator. Whereas the modified within estimator drops the partially observed cross-sectional units, the modified Chamberlain estimator still uses both time periods of covariate data for such units. Mundlak estimator: When s it = 0 observations are dropped, the resulting pooled OLS estimator y it on x it, 1, x i ) for s it consistent. 5 = 1 observations) is no longer guaranteed to be Interestingly, the modified Chamberlain estimator is able to incorporate cross-sectional units for which only a single y it value is available. This particular feature seems to be new to the literature on fixed-effects models, although Altonji and Matzkin 2005) have previously considered estimation with a single y it and multiple covariate observations) in a nonparametric panel-data model with heterogeneity but not of a pure fixed-effects form). 6 The modified Chamberlain estimator provides efficiency gains since it effectively uses more observations than the within estimator. The modified Chamberlain estimator includes one of time periods for the partially observed cross-sectional units within the pooled OLS estimation, whereas the within estimator is equivalent to a pooled OLS on only the fully observed cross-sectional units. To illustrate the efficiency gains from the modified Chamberlain estimator as compared to the modified within estimator), we report the results of some simple simulations. Using the following design for n = 1000 and T = 2, x 1, x 2, u 1, u 2 N0, 1), a = 1 + 2x 1, y t = x t + a + u t, and y it missing at random three simulations: 0% missingness, 20% missingness, 40% missingness), we report the results from a representative simulation in Table 1. For this design, 5 Of course, if the partially observed cross-sectional units were dropped entirely, the resulting estimator would remain equivalent to the within estimator. 6 Specifically, Altonji and Matzkin 2005) require the heterogeneity to depend upon some function of the covariates like x i ). [5]

7 Table 1: Simulation results with missing y it values None missing 20% y 2 missing 40% y 2 missing Baseline within on full data) ) ) ) Within on partial data ) ) ) Chamberlain ) ) ) Mundlak ) ) ) Coefficient estimates with standard errors in parentheses) are reported. Sample size n = The simulation design is described in the text. the Chamberlain approach eliminates roughly half of the inefficiency of the within estimator i.e., comparing the within on partial data to the baseline on the full data). 3 GMM estimation with missing dependent variables In this section, we consider GMM estimation of the fixed-effects model based upon the projection form in equation 2.3). Under Assumption 1, the composite error disturbance a i + u it ) is uncorrelated with the covariates from all time periods i.e., Ex isa i + u it )) = 0 for all s, t {1,..., T }). Moreover, under Assumption 2, these orthogonality properties remain true even when conditioning on observability s it = 1). The following moment functions correspond to these orthogonality conditions: s it 1 x i ) y it x it β ψ x i1 λ 1 x i2 λ 2 x it λ T ) for t = 1,..., T. 3.4) Recall that x i is the kt 1 stacked vector of covariates from all time periods. Let gz i, θ) denote the stacked vector of all the moment functions in 3.4), where z i y i, x i, s i ). There are a total of T kt + 1) moments conditions and T + 1)k + 1 parameters, meaning there are T 2 T 1)k + T 1) overidentifying restrictions. The orthogonality conditions imply that the moment functions have expectation zero at the true parameter values. For identification i.e., non-zero expectation at other parameter values), we need the following additional full-rank condition: [6]

8 Assumption 3 Full rank) EV i V i ) has full column rank, where V i s i1 x i1 s i1 s i1 x i s i2 x i2 s i2 s i2 x i... s it x it s it s it x i Note that a given row of V i matrix is the row of RHS variables x it, 1, x i) for the Chamberlain regression when s it = 1 y it observed) and is a row of zeros when s it = 0 y it missing). The full-rank condition, aside from ruling out the usual linear dependence of covariates, also rules out a situation in which a given time period s dependent variable for example, y i1 ) is missing for all i. The following lemma provides the GMM identification result: Lemma 1 Under Assumptions 1 and 2, E[gz i, θ 0 )] = 0. If Assumption 3 also holds, then E[gz i, θ)] 0 for θ θ 0. Let ˆθ denote the unweighted GMM estimator obtained by minimizing n 1 n gz i, θ) ) n 1 n. gz i, θ) ). 3.5) The GMM estimator can be implemented without numerical optimization by using instrumentalvariables methods. Specifically, define the instrumental-variable matrix Z i as s i1 s i1 x i s i2 s i2 x i 0 0. Z i s it s it x i 3.6) which corresponds to using 1 x i) as instruments in each time period for which y it is observed. The Z i instrument matrix is T kt + 1). Then, the unweighted GMM estimator ˆθ see equation 3.5)) can be obtained directly as the system IV estimator ˆθ = V ZZ V ) 1 V ZZ Y ), 3.7) where V, Z, and Y are the stacked versions of V i, Z i, and y i with dimensions nt T + 1)k + 1), nt T kt + 1), and nt 1, respectively). [7]

9 The unweighted GMM estimator is not efficient. The GMM objective function that allows for weighting is n 1 n gz i, θ)) Ŵ The system 2SLS estimator has Ŵ2SLS = Z Z n Note that ˆθ 2SLS n 1 n ) 1, so that gz i, θ) ˆθ2SLS = V ZŴ2SLSZ V ) 1 V ZŴ2SLSZ Y ). ). 3.8) is numerically equivalent to the modified Chamberlain linear-regression estimator described in Section 2.1, specifically the pooled OLS estimator of y it on 1, x it, x i) for the subsample of s it = 1 observations. 7 To obtain the optimal GMM estimator, the standard two-step approach can be used. For instance, after obtaining the system 2SLS estimator ˆθ 2SLS, the optimal GMM estimator θ minimizes the objective function 4.13) where the optimal weighting matrix Ŵ is given by Ŵ = n 1 n gz i, ˆθ 2SLS )gz i, ˆθ 2SLS ) ) 1. The optimal GMM estimator can be obtained directly as θ = V ZŴ Z V ) 1 V ZŴ Z Y ). 4 Exchangeable panels In this section, we consider a fixed-effects model in which the time periods are exchangeable. The notion of exchangeability in this context means that the fixed effect is independent of the ordering of the time periods in the data. Examples could include panel data for which t does not actually index time, such as sibling data, twins data, etc. Even in such cases, however, the exchangeability assumption will rule out an order effect, such as a birth-order effect for siblings data.) The exchangeability assumption is defined as follows: Assumption 4 Exchangeability) Dc i x i1 = x 1, x i2 = x 2,..., x it = x T ) = Dc i x ij1 = x 1, x ij2 = x 2,..., x ijt = x T ) for any x 1, x 2,..., x T ) and any permutation of the set {1, 2,..., T } denoted j 1, j 2,..., j T ). 7 To see this equivalence, note that the first-stage regression is vacuous in that the fitted values are identical to the pooled-ols covariate matrix specifically, ZZ Z) 1 Z V = V. [8]

10 Recalling the projection of the fixed effect c i onto the covariates, c i = ψ 0 + x i1 λ 01 + x i2 λ x it λ 0T + a i, Assumption 4 implies that the λ coefficient vectors must be identical for each time period that is, λ 01 = λ 02 = = λ 0T = λ 0 for some k 1 vector λ 0. Plugging these equalities back into the projection equation yields or equivalently T ) c i = ψ 0 + x it λ 0 + a i, 4.9) t=1 c i = ψ 0 + x i γ 0 + a i, where γ 0 = T λ ) Importantly, the projection error a i in 4.9) and 4.10) is still uncorrelated with the covariates x i1, x i2,..., x it ). In general, this orthogonality does not hold without exchangeability when one projects c i upon x i. The GMM estimator can incorporate the exchangeability restriction by writing the orthogonality conditions using the projection equation 4.10) and conditioning on observability, E x i a i + u it ) s i ) = E x i y it x it β 0 ψ 0 x i γ 0 ) s i ) = ) The vector of moment functions, denoted hz i, θ), corresponding to these orthogonality conditions is obtained by stacking the following functions s it 1 x i ) y it x it β ψ x i γ) for t = 1,..., T. 4.12) There are T kt + 1) moment functions, as in the non-exchangeable case of Section 3, but now only 2k + 1 parameters. The associated GMM objective function is n 1 n hz i, θ)) Ŵ n 1 n hz i, θ) ) 4.13) for a weighting matrix Ŵ. As in the non-exchangeable case, the GMM estimator can be obtained directly using IV methods. The instrument matrix Z i and its stacked version [9]

11 Z) remains the same since the instruments in 4.12) are unchanged. The covariate matrix denoted Vi e ) is different in the exchangeable case, as x i is used in place of the full x i vector, s i1 x i1 s i1 s i1 x i X i e s i2 x i2 s i2 s i2 x i ) s it x it s it s it x i The covariate matrix V e i has dimension nt 2k + 1). The form of the IV estimator, depending upon the weighting matrix Ŵ, is V e ZŴ Z V e ) 1 V e ZŴ Z Y ), 4.15) where V e is the stacked version of Vi e. The system 2SLS estimator, denoted ˆθ e 2SLS with Ŵ = ) Z Z 1, n remains numerically equivalent to the modified Chamberlain linear-regression estimator of Section 2.1. The optimal GMM estimator can be obtained by using the optimal weighting matrix Ŵ = n 1 in the closed-form expression 4.15). n hz i, ˆθ 2SLS )hz i, ˆθ 2SLS ) ) 1 An interesting feature of the model with exchangeability is that the parameters can be identified even when only a single y it is available for each i. As an example, if one had data on twins with y i1 observed, y i2 missing, and both x i1 and x i2 observed, the leastsquares regression of y i1 on x i1, 1, x i ) would consistently estimate β 0, ψ 0, γ 0 ). 8 In terms of the previous full-rank condition Assumption 3) discussed in Section 3, the model with exchangeability requires that EVi e Vi e ) have full column rank. If only y i1 is observed s i1 = 1, s i2 = = s it = 0), this full-rank condition is equivalent to having full rank of EV e i1 V e i1) where V e i1 = [x i1 1 x i ] is the first row of V e i. This condition will hold with time-varying x it, whereas the analogous condition for the non-exchangeable case would not; in the nonexchangeable case, recall that the first row of V i was [x i1 1 x i1 x it ], for which x i1 causes colinearity problems. 8 This idea was previously discussed by Altonji and Matzkin 2005, p. 1066), but their discussion starts from a slightly different model their equation 2.9)) under additional restrictions. In contrast, the discussion here points out that the exchangeability assumption yields this identification result for the linear fixed-effects model without any further restrictions. [10]

12 If Assumption 4 is true, the optimal GMM estimator above will be at least as efficient as the optimal GMM estimator of Section 3 since it incorporates the restrictions on the λ parameters. Note that one could directly test an implication of exchangeability by using any consistent estimator from Section 2 and testing the null hypothesis H 0 : λ 01 = λ 02 = = λ 0T. 5 Lagged dependent variables In this section, we consider estimation in a fixed-effects model where a lagged dependent variable enters as an explanatory variable often called an autoregressive fixed-effects model). Missingness of y it causes further complications here since the missingness now affects both the dependent and independent variables of the model. To focus ideas, we consider a model with no additional covariates x it, although inclusion of such covariates would be straightforward: 9 y it = ρ 0 y i,t 1 + c i + u it i = 1,..., n; t = 1,..., T ). 5.16) where Eu it y i,t 1,..., y i0, c i, s i ) = 0 for t = 1,..., T. 5.17) The condition on the disturbances in 5.17), which implies that the AR1) structure completely captures the dynamics, is standard for this model see, e.g., Alonso-Borrego and Arellano 1999)). As in the preceding sections s it = 1 if y it is observed and s it = 0 if y it is missing. The initial condition is given by y i0. Without loss of generality, we will assume that y i0 is never missing. 10 Consider the projection of the fixed effect c i onto the initial condition, given by 11 c i = ψ0 + γ0y i0 + a i, Ea i y i0) = ) 9 The projection described below would include x i1,..., x it ). 10 If y i0 is missing, the first non-missing y it can be treated as the initial condition under the missingness assumption below). 11 The idea of projecting the fixed effect onto the initial condition was first considered by Chamberlain 1980), although his original treatment made the additional assumption of joint normality of the u it disturbances conditional on y i0 ). [11]

13 For the projection-based GMM estimator to be consistent, we require a condition analogous to Assumption 2, specifically Ea i y i0 s i ) = 0 here. In order to construct the GMM estimator that allows for missingness, re-write y it as a function of the initial condition y i0 by recursively applying equation 5.16) and plugging in the projection 5.18). For instance, y i1 can be expressed in terms of y i0 as follows: y i1 = ρ 0 y i0 + c i + u i1 = ρ 0 y i0 + ψ 0 + γ 0 y i0 + a i + u i1 = ψ 0 + ρ 0 + γ 0 )y i0 + a i + u i1. The next observation, y i2, can then be written y i2 = ρ 0 y i1 + c i + u i2 = ρρ 0 y i0 + ψ 0 + γ 0 y i0 + a i + u i1 ) + ψ 0 + γ 0 y i0 + a i + u i2 = ψ ρ 0 ) + ρ γ ρ 0 ))y i0 + a i + ρ 0 u i1 + u i2. Repeating the recursive substitution yields the general formula ) t 1 y it = ψ 0 ρ s 0 + ρ t0 t 1 t 1 + γ 0 y i0 + a i + ρ s 0u i,t s. 5.19) s=0 ρ s 0 s=0 The moment conditions are based upon the composite error in 5.19), a i + t 1 s=0 ρ s 0u i,t s, being uncorrelated with y i0 conditional on s i ). Specifically, the vector of moment functions would be obtained by stacking the following functions ) ) ) 1 t 1 s it y y it ψ ρ s ρ t t 1 + γ ρ s y i0 for t = 1,..., T. 5.20) i0 s=0 GMM estimation proceeds in the usual way. In this situation without covariates), there are a total of 2T moments and 3 parameters ρ 0, φ 0, γ 0 ) being estimated. For T = 2 a maximum of three observed y it values), identification requires that the probabilities of observability for y i1 and y i2 are non-zero. Interestingly, the model parameters can be identified when all crosssectional units have only two observed y it values for example, if y i0, y i1 ) are observed for some observations and y i0, y i2 ) are observed for the remaining observations. That is, we require three time periods of data as in the fully observed case) but obtain identification with only two dependent variables for each cross-sectional unit. [12] s=0 s=0

14 References Alonso-Borrego, Cesar and Manuel Arellano, 1999, Symmetrically normalized instrumentalvariable estimation using panel data, Journal of Business and Economic Statistics 17, Altonji, Joseph and Rosa L. Matzkin, 2005, Cross section and panel data nonseparable models with endogenous regressors, Econometrica 73, Chamberlain, Gary, 1980, Analysis of covariance with qualitative data, Review of Economic Studies 47, Chamberlain, Gary, 1982, Multivariate regression models for panel data, Journal of Econometrics 18, Mundlak, Yair, 1978, On the pooling of time series and cross sectional data, Econometrica 56, Wooldridge, Jeffrey M., 2002, Econometric analysis of cross section and panel data, MIT Press. [13]

A Course in Applied Econometrics Lecture 4: Linear Panel Data Models, II. Jeff Wooldridge IRP Lectures, UW Madison, August 2008

A Course in Applied Econometrics Lecture 4: Linear Panel Data Models, II. Jeff Wooldridge IRP Lectures, UW Madison, August 2008 A Course in Applied Econometrics Lecture 4: Linear Panel Data Models, II Jeff Wooldridge IRP Lectures, UW Madison, August 2008 5. Estimating Production Functions Using Proxy Variables 6. Pseudo Panels

More information

When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Longitudinal Data?

When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Longitudinal Data? When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Longitudinal Data? Kosuke Imai Department of Politics Center for Statistics and Machine Learning Princeton University

More information

Econometrics of Panel Data

Econometrics of Panel Data Econometrics of Panel Data Jakub Mućk Meeting # 6 Jakub Mućk Econometrics of Panel Data Meeting # 6 1 / 36 Outline 1 The First-Difference (FD) estimator 2 Dynamic panel data models 3 The Anderson and Hsiao

More information

Non-linear panel data modeling

Non-linear panel data modeling Non-linear panel data modeling Laura Magazzini University of Verona laura.magazzini@univr.it http://dse.univr.it/magazzini May 2010 Laura Magazzini (@univr.it) Non-linear panel data modeling May 2010 1

More information

A Course in Applied Econometrics Lecture 14: Control Functions and Related Methods. Jeff Wooldridge IRP Lectures, UW Madison, August 2008

A Course in Applied Econometrics Lecture 14: Control Functions and Related Methods. Jeff Wooldridge IRP Lectures, UW Madison, August 2008 A Course in Applied Econometrics Lecture 14: Control Functions and Related Methods Jeff Wooldridge IRP Lectures, UW Madison, August 2008 1. Linear-in-Parameters Models: IV versus Control Functions 2. Correlated

More information

CRE METHODS FOR UNBALANCED PANELS Correlated Random Effects Panel Data Models IZA Summer School in Labor Economics May 13-19, 2013 Jeffrey M.

CRE METHODS FOR UNBALANCED PANELS Correlated Random Effects Panel Data Models IZA Summer School in Labor Economics May 13-19, 2013 Jeffrey M. CRE METHODS FOR UNBALANCED PANELS Correlated Random Effects Panel Data Models IZA Summer School in Labor Economics May 13-19, 2013 Jeffrey M. Wooldridge Michigan State University 1. Introduction 2. Linear

More information

Estimation of partial effects in non-linear panel data models

Estimation of partial effects in non-linear panel data models Estimation of partial effects in non-linear panel data models by Jason Abrevaya and Yu-Chin Hsu This version: September 2011 ABSTRACT Nonlinearity and heterogeneity complicate the estimation and interpretation

More information

Specification testing in panel data models estimated by fixed effects with instrumental variables

Specification testing in panel data models estimated by fixed effects with instrumental variables Specification testing in panel data models estimated by fixed effects wh instrumental variables Carrie Falls Department of Economics Michigan State Universy Abstract I show that a handful of the regressions

More information

Econometric Analysis of Cross Section and Panel Data

Econometric Analysis of Cross Section and Panel Data Econometric Analysis of Cross Section and Panel Data Jeffrey M. Wooldridge / The MIT Press Cambridge, Massachusetts London, England Contents Preface Acknowledgments xvii xxiii I INTRODUCTION AND BACKGROUND

More information

Chapter 2. Dynamic panel data models

Chapter 2. Dynamic panel data models Chapter 2. Dynamic panel data models School of Economics and Management - University of Geneva Christophe Hurlin, Université of Orléans University of Orléans April 2018 C. Hurlin (University of Orléans)

More information

Econometrics. Week 6. Fall Institute of Economic Studies Faculty of Social Sciences Charles University in Prague

Econometrics. Week 6. Fall Institute of Economic Studies Faculty of Social Sciences Charles University in Prague Econometrics Week 6 Institute of Economic Studies Faculty of Social Sciences Charles University in Prague Fall 2012 1 / 21 Recommended Reading For the today Advanced Panel Data Methods. Chapter 14 (pp.

More information

Panel Data Models. Chapter 5. Financial Econometrics. Michael Hauser WS17/18 1 / 63

Panel Data Models. Chapter 5. Financial Econometrics. Michael Hauser WS17/18 1 / 63 1 / 63 Panel Data Models Chapter 5 Financial Econometrics Michael Hauser WS17/18 2 / 63 Content Data structures: Times series, cross sectional, panel data, pooled data Static linear panel data models:

More information

When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Panel Data?

When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Panel Data? When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Panel Data? Kosuke Imai Department of Politics Center for Statistics and Machine Learning Princeton University Joint

More information

Efficiency of repeated-cross-section estimators in fixed-effects models

Efficiency of repeated-cross-section estimators in fixed-effects models Efficiency of repeated-cross-section estimators in fixed-effects models Montezuma Dumangane and Nicoletta Rosati CEMAPRE and ISEG-UTL January 2009 Abstract PRELIMINARY AND INCOMPLETE Exploiting across

More information

1 Estimation of Persistent Dynamic Panel Data. Motivation

1 Estimation of Persistent Dynamic Panel Data. Motivation 1 Estimation of Persistent Dynamic Panel Data. Motivation Consider the following Dynamic Panel Data (DPD) model y it = y it 1 ρ + x it β + µ i + v it (1.1) with i = {1, 2,..., N} denoting the individual

More information

Specification Test for Instrumental Variables Regression with Many Instruments

Specification Test for Instrumental Variables Regression with Many Instruments Specification Test for Instrumental Variables Regression with Many Instruments Yoonseok Lee and Ryo Okui April 009 Preliminary; comments are welcome Abstract This paper considers specification testing

More information

Spatial Regression. 13. Spatial Panels (1) Luc Anselin. Copyright 2017 by Luc Anselin, All Rights Reserved

Spatial Regression. 13. Spatial Panels (1) Luc Anselin.  Copyright 2017 by Luc Anselin, All Rights Reserved Spatial Regression 13. Spatial Panels (1) Luc Anselin http://spatial.uchicago.edu 1 basic concepts dynamic panels pooled spatial panels 2 Basic Concepts 3 Data Structures 4 Two-Dimensional Data cross-section/space

More information

Final Exam. Economics 835: Econometrics. Fall 2010

Final Exam. Economics 835: Econometrics. Fall 2010 Final Exam Economics 835: Econometrics Fall 2010 Please answer the question I ask - no more and no less - and remember that the correct answer is often short and simple. 1 Some short questions a) For each

More information

Multiple Equation GMM with Common Coefficients: Panel Data

Multiple Equation GMM with Common Coefficients: Panel Data Multiple Equation GMM with Common Coefficients: Panel Data Eric Zivot Winter 2013 Multi-equation GMM with common coefficients Example (panel wage equation) 69 = + 69 + + 69 + 1 80 = + 80 + + 80 + 2 Note:

More information

A Course in Applied Econometrics Lecture 18: Missing Data. Jeff Wooldridge IRP Lectures, UW Madison, August Linear model with IVs: y i x i u i,

A Course in Applied Econometrics Lecture 18: Missing Data. Jeff Wooldridge IRP Lectures, UW Madison, August Linear model with IVs: y i x i u i, A Course in Applied Econometrics Lecture 18: Missing Data Jeff Wooldridge IRP Lectures, UW Madison, August 2008 1. When Can Missing Data be Ignored? 2. Inverse Probability Weighting 3. Imputation 4. Heckman-Type

More information

Quantile Regression for Panel Data Models with Fixed Effects and Small T : Identification and Estimation

Quantile Regression for Panel Data Models with Fixed Effects and Small T : Identification and Estimation Quantile Regression for Panel Data Models with Fixed Effects and Small T : Identification and Estimation Maria Ponomareva University of Western Ontario May 8, 2011 Abstract This paper proposes a moments-based

More information

Short T Panels - Review

Short T Panels - Review Short T Panels - Review We have looked at methods for estimating parameters on time-varying explanatory variables consistently in panels with many cross-section observation units but a small number of

More information

A Robust Approach to Estimating Production Functions: Replication of the ACF procedure

A Robust Approach to Estimating Production Functions: Replication of the ACF procedure A Robust Approach to Estimating Production Functions: Replication of the ACF procedure Kyoo il Kim Michigan State University Yao Luo University of Toronto Yingjun Su IESR, Jinan University August 2018

More information

Panel Data Seminar. Discrete Response Models. Crest-Insee. 11 April 2008

Panel Data Seminar. Discrete Response Models. Crest-Insee. 11 April 2008 Panel Data Seminar Discrete Response Models Romain Aeberhardt Laurent Davezies Crest-Insee 11 April 2008 Aeberhardt and Davezies (Crest-Insee) Panel Data Seminar 11 April 2008 1 / 29 Contents Overview

More information

Economics 536 Lecture 7. Introduction to Specification Testing in Dynamic Econometric Models

Economics 536 Lecture 7. Introduction to Specification Testing in Dynamic Econometric Models University of Illinois Fall 2016 Department of Economics Roger Koenker Economics 536 Lecture 7 Introduction to Specification Testing in Dynamic Econometric Models In this lecture I want to briefly describe

More information

Econometrics II - EXAM Answer each question in separate sheets in three hours

Econometrics II - EXAM Answer each question in separate sheets in three hours Econometrics II - EXAM Answer each question in separate sheets in three hours. Let u and u be jointly Gaussian and independent of z in all the equations. a Investigate the identification of the following

More information

Chapter 6. Panel Data. Joan Llull. Quantitative Statistical Methods II Barcelona GSE

Chapter 6. Panel Data. Joan Llull. Quantitative Statistical Methods II Barcelona GSE Chapter 6. Panel Data Joan Llull Quantitative Statistical Methods II Barcelona GSE Introduction Chapter 6. Panel Data 2 Panel data The term panel data refers to data sets with repeated observations over

More information

Econometrics of Panel Data

Econometrics of Panel Data Econometrics of Panel Data Jakub Mućk Meeting # 2 Jakub Mućk Econometrics of Panel Data Meeting # 2 1 / 26 Outline 1 Fixed effects model The Least Squares Dummy Variable Estimator The Fixed Effect (Within

More information

Bias Correction Methods for Dynamic Panel Data Models with Fixed Effects

Bias Correction Methods for Dynamic Panel Data Models with Fixed Effects MPRA Munich Personal RePEc Archive Bias Correction Methods for Dynamic Panel Data Models with Fixed Effects Mohamed R. Abonazel Department of Applied Statistics and Econometrics, Institute of Statistical

More information

Specification Tests in Unbalanced Panels with Endogeneity.

Specification Tests in Unbalanced Panels with Endogeneity. Specification Tests in Unbalanced Panels with Endogeneity. Riju Joshi Jeffrey M. Wooldridge June 22, 2017 Abstract This paper develops specification tests for unbalanced panels with endogenous explanatory

More information

ECONOMETRICS II (ECO 2401S) University of Toronto. Department of Economics. Spring 2013 Instructor: Victor Aguirregabiria

ECONOMETRICS II (ECO 2401S) University of Toronto. Department of Economics. Spring 2013 Instructor: Victor Aguirregabiria ECONOMETRICS II (ECO 2401S) University of Toronto. Department of Economics. Spring 2013 Instructor: Victor Aguirregabiria SOLUTION TO FINAL EXAM Friday, April 12, 2013. From 9:00-12:00 (3 hours) INSTRUCTIONS:

More information

Econometrics. Week 4. Fall Institute of Economic Studies Faculty of Social Sciences Charles University in Prague

Econometrics. Week 4. Fall Institute of Economic Studies Faculty of Social Sciences Charles University in Prague Econometrics Week 4 Institute of Economic Studies Faculty of Social Sciences Charles University in Prague Fall 2012 1 / 23 Recommended Reading For the today Serial correlation and heteroskedasticity in

More information

10 Panel Data. Andrius Buteikis,

10 Panel Data. Andrius Buteikis, 10 Panel Data Andrius Buteikis, andrius.buteikis@mif.vu.lt http://web.vu.lt/mif/a.buteikis/ Introduction Panel data combines cross-sectional and time series data: the same individuals (persons, firms,

More information

Linear models. Linear models are computationally convenient and remain widely used in. applied econometric research

Linear models. Linear models are computationally convenient and remain widely used in. applied econometric research Linear models Linear models are computationally convenient and remain widely used in applied econometric research Our main focus in these lectures will be on single equation linear models of the form y

More information

Advanced Econometrics

Advanced Econometrics Based on the textbook by Verbeek: A Guide to Modern Econometrics Robert M. Kunst robert.kunst@univie.ac.at University of Vienna and Institute for Advanced Studies Vienna May 16, 2013 Outline Univariate

More information

Flexible Estimation of Treatment Effect Parameters

Flexible Estimation of Treatment Effect Parameters Flexible Estimation of Treatment Effect Parameters Thomas MaCurdy a and Xiaohong Chen b and Han Hong c Introduction Many empirical studies of program evaluations are complicated by the presence of both

More information

Estimation of Dynamic Panel Data Models with Sample Selection

Estimation of Dynamic Panel Data Models with Sample Selection === Estimation of Dynamic Panel Data Models with Sample Selection Anastasia Semykina* Department of Economics Florida State University Tallahassee, FL 32306-2180 asemykina@fsu.edu Jeffrey M. Wooldridge

More information

Instrumental Variables and the Problem of Endogeneity

Instrumental Variables and the Problem of Endogeneity Instrumental Variables and the Problem of Endogeneity September 15, 2015 1 / 38 Exogeneity: Important Assumption of OLS In a standard OLS framework, y = xβ + ɛ (1) and for unbiasedness we need E[x ɛ] =

More information

IDENTIFICATION OF MARGINAL EFFECTS IN NONSEPARABLE MODELS WITHOUT MONOTONICITY

IDENTIFICATION OF MARGINAL EFFECTS IN NONSEPARABLE MODELS WITHOUT MONOTONICITY Econometrica, Vol. 75, No. 5 (September, 2007), 1513 1518 IDENTIFICATION OF MARGINAL EFFECTS IN NONSEPARABLE MODELS WITHOUT MONOTONICITY BY STEFAN HODERLEIN AND ENNO MAMMEN 1 Nonseparable models do not

More information

EC327: Advanced Econometrics, Spring 2007

EC327: Advanced Econometrics, Spring 2007 EC327: Advanced Econometrics, Spring 2007 Wooldridge, Introductory Econometrics (3rd ed, 2006) Chapter 14: Advanced panel data methods Fixed effects estimators We discussed the first difference (FD) model

More information

Econometrics Master in Business and Quantitative Methods

Econometrics Master in Business and Quantitative Methods Econometrics Master in Business and Quantitative Methods Helena Veiga Universidad Carlos III de Madrid Models with discrete dependent variables and applications of panel data methods in all fields of economics

More information

Estimating Panel Data Models in the Presence of Endogeneity and Selection

Estimating Panel Data Models in the Presence of Endogeneity and Selection ================ Estimating Panel Data Models in the Presence of Endogeneity and Selection Anastasia Semykina Department of Economics Florida State University Tallahassee, FL 32306-2180 asemykina@fsu.edu

More information

Linear Models in Econometrics

Linear Models in Econometrics Linear Models in Econometrics Nicky Grant At the most fundamental level econometrics is the development of statistical techniques suited primarily to answering economic questions and testing economic theories.

More information

Time Invariant Variables and Panel Data Models : A Generalised Frisch- Waugh Theorem and its Implications

Time Invariant Variables and Panel Data Models : A Generalised Frisch- Waugh Theorem and its Implications Time Invariant Variables and Panel Data Models : A Generalised Frisch- Waugh Theorem and its Implications Jaya Krishnakumar No 2004.01 Cahiers du département d économétrie Faculté des sciences économiques

More information

Econometrics I KS. Module 2: Multivariate Linear Regression. Alexander Ahammer. This version: April 16, 2018

Econometrics I KS. Module 2: Multivariate Linear Regression. Alexander Ahammer. This version: April 16, 2018 Econometrics I KS Module 2: Multivariate Linear Regression Alexander Ahammer Department of Economics Johannes Kepler University of Linz This version: April 16, 2018 Alexander Ahammer (JKU) Module 2: Multivariate

More information

Repeated observations on the same cross-section of individual units. Important advantages relative to pure cross-section data

Repeated observations on the same cross-section of individual units. Important advantages relative to pure cross-section data Panel data Repeated observations on the same cross-section of individual units. Important advantages relative to pure cross-section data - possible to control for some unobserved heterogeneity - possible

More information

Econ 582 Fixed Effects Estimation of Panel Data

Econ 582 Fixed Effects Estimation of Panel Data Econ 582 Fixed Effects Estimation of Panel Data Eric Zivot May 28, 2012 Panel Data Framework = x 0 β + = 1 (individuals); =1 (time periods) y 1 = X β ( ) ( 1) + ε Main question: Is x uncorrelated with?

More information

On IV estimation of the dynamic binary panel data model with fixed effects

On IV estimation of the dynamic binary panel data model with fixed effects On IV estimation of the dynamic binary panel data model with fixed effects Andrew Adrian Yu Pua March 30, 2015 Abstract A big part of applied research still uses IV to estimate a dynamic linear probability

More information

Efficient Estimation of Dynamic Panel Data Models: Alternative Assumptions and Simplified Estimation

Efficient Estimation of Dynamic Panel Data Models: Alternative Assumptions and Simplified Estimation Efficient Estimation of Dynamic Panel Data Models: Alternative Assumptions and Simplified Estimation Seung C. Ahn Arizona State University, Tempe, AZ 85187, USA Peter Schmidt * Michigan State University,

More information

A GMM approach for dealing with missing data on regressors and instruments

A GMM approach for dealing with missing data on regressors and instruments A GMM approach for dealing with missing data on regressors and instruments Jason Abrevaya Department of Economics University of Texas Stephen G. Donald Department of Economics University of Texas This

More information

New Developments in Econometrics Lecture 16: Quantile Estimation

New Developments in Econometrics Lecture 16: Quantile Estimation New Developments in Econometrics Lecture 16: Quantile Estimation Jeff Wooldridge Cemmap Lectures, UCL, June 2009 1. Review of Means, Medians, and Quantiles 2. Some Useful Asymptotic Results 3. Quantile

More information

Applied Health Economics (for B.Sc.)

Applied Health Economics (for B.Sc.) Applied Health Economics (for B.Sc.) Helmut Farbmacher Department of Economics University of Mannheim Autumn Semester 2017 Outlook 1 Linear models (OLS, Omitted variables, 2SLS) 2 Limited and qualitative

More information

Instrumental Variables

Instrumental Variables Università di Pavia 2010 Instrumental Variables Eduardo Rossi Exogeneity Exogeneity Assumption: the explanatory variables which form the columns of X are exogenous. It implies that any randomness in the

More information

PANEL DATA RANDOM AND FIXED EFFECTS MODEL. Professor Menelaos Karanasos. December Panel Data (Institute) PANEL DATA December / 1

PANEL DATA RANDOM AND FIXED EFFECTS MODEL. Professor Menelaos Karanasos. December Panel Data (Institute) PANEL DATA December / 1 PANEL DATA RANDOM AND FIXED EFFECTS MODEL Professor Menelaos Karanasos December 2011 PANEL DATA Notation y it is the value of the dependent variable for cross-section unit i at time t where i = 1,...,

More information

Dynamic Panels. Chapter Introduction Autoregressive Model

Dynamic Panels. Chapter Introduction Autoregressive Model Chapter 11 Dynamic Panels This chapter covers the econometrics methods to estimate dynamic panel data models, and presents examples in Stata to illustrate the use of these procedures. The topics in this

More information

Review of Classical Least Squares. James L. Powell Department of Economics University of California, Berkeley

Review of Classical Least Squares. James L. Powell Department of Economics University of California, Berkeley Review of Classical Least Squares James L. Powell Department of Economics University of California, Berkeley The Classical Linear Model The object of least squares regression methods is to model and estimate

More information

1 Motivation for Instrumental Variable (IV) Regression

1 Motivation for Instrumental Variable (IV) Regression ECON 370: IV & 2SLS 1 Instrumental Variables Estimation and Two Stage Least Squares Econometric Methods, ECON 370 Let s get back to the thiking in terms of cross sectional (or pooled cross sectional) data

More information

Lecture: Simultaneous Equation Model (Wooldridge s Book Chapter 16)

Lecture: Simultaneous Equation Model (Wooldridge s Book Chapter 16) Lecture: Simultaneous Equation Model (Wooldridge s Book Chapter 16) 1 2 Model Consider a system of two regressions y 1 = β 1 y 2 + u 1 (1) y 2 = β 2 y 1 + u 2 (2) This is a simultaneous equation model

More information

Economic modelling and forecasting

Economic modelling and forecasting Economic modelling and forecasting 2-6 February 2015 Bank of England he generalised method of moments Ole Rummel Adviser, CCBS at the Bank of England ole.rummel@bankofengland.co.uk Outline Classical estimation

More information

Wooldridge, Introductory Econometrics, 4th ed. Chapter 15: Instrumental variables and two stage least squares

Wooldridge, Introductory Econometrics, 4th ed. Chapter 15: Instrumental variables and two stage least squares Wooldridge, Introductory Econometrics, 4th ed. Chapter 15: Instrumental variables and two stage least squares Many economic models involve endogeneity: that is, a theoretical relationship does not fit

More information

EFFICIENT ESTIMATION USING PANEL DATA 1. INTRODUCTION

EFFICIENT ESTIMATION USING PANEL DATA 1. INTRODUCTION Econornetrica, Vol. 57, No. 3 (May, 1989), 695-700 EFFICIENT ESTIMATION USING PANEL DATA BY TREVOR S. BREUSCH, GRAYHAM E. MIZON, AND PETER SCHMIDT' 1. INTRODUCTION IN AN IMPORTANT RECENT PAPER, Hausman

More information

Generalized Method of Moments: I. Chapter 9, R. Davidson and J.G. MacKinnon, Econometric Theory and Methods, 2004, Oxford.

Generalized Method of Moments: I. Chapter 9, R. Davidson and J.G. MacKinnon, Econometric Theory and Methods, 2004, Oxford. Generalized Method of Moments: I References Chapter 9, R. Davidson and J.G. MacKinnon, Econometric heory and Methods, 2004, Oxford. Chapter 5, B. E. Hansen, Econometrics, 2006. http://www.ssc.wisc.edu/~bhansen/notes/notes.htm

More information

When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Longitudinal Data?

When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Longitudinal Data? When Should We Use Linear Fixed Effects Regression Models for Causal Inference with Longitudinal Data? Kosuke Imai Princeton University Asian Political Methodology Conference University of Sydney Joint

More information

Econometric Methods and Applications II Chapter 2: Simultaneous equations. Econometric Methods and Applications II, Chapter 2, Slide 1

Econometric Methods and Applications II Chapter 2: Simultaneous equations. Econometric Methods and Applications II, Chapter 2, Slide 1 Econometric Methods and Applications II Chapter 2: Simultaneous equations Econometric Methods and Applications II, Chapter 2, Slide 1 2.1 Introduction An example motivating the problem of simultaneous

More information

Linear dynamic panel data models

Linear dynamic panel data models Linear dynamic panel data models Laura Magazzini University of Verona L. Magazzini (UniVR) Dynamic PD 1 / 67 Linear dynamic panel data models Dynamic panel data models Notation & Assumptions One of the

More information

Questions and Answers on Unit Roots, Cointegration, VARs and VECMs

Questions and Answers on Unit Roots, Cointegration, VARs and VECMs Questions and Answers on Unit Roots, Cointegration, VARs and VECMs L. Magee Winter, 2012 1. Let ɛ t, t = 1,..., T be a series of independent draws from a N[0,1] distribution. Let w t, t = 1,..., T, be

More information

Notes on Panel Data and Fixed Effects models

Notes on Panel Data and Fixed Effects models Notes on Panel Data and Fixed Effects models Michele Pellizzari IGIER-Bocconi, IZA and frdb These notes are based on a combination of the treatment of panel data in three books: (i) Arellano M 2003 Panel

More information

Lecture 6: Dynamic panel models 1

Lecture 6: Dynamic panel models 1 Lecture 6: Dynamic panel models 1 Ragnar Nymoen Department of Economics, UiO 16 February 2010 Main issues and references Pre-determinedness and endogeneity of lagged regressors in FE model, and RE model

More information

Panel Data Model (January 9, 2018)

Panel Data Model (January 9, 2018) Ch 11 Panel Data Model (January 9, 2018) 1 Introduction Data sets that combine time series and cross sections are common in econometrics For example, the published statistics of the OECD contain numerous

More information

Generated Covariates in Nonparametric Estimation: A Short Review.

Generated Covariates in Nonparametric Estimation: A Short Review. Generated Covariates in Nonparametric Estimation: A Short Review. Enno Mammen, Christoph Rothe, and Melanie Schienle Abstract In many applications, covariates are not observed but have to be estimated

More information

1. You have data on years of work experience, EXPER, its square, EXPER2, years of education, EDUC, and the log of hourly wages, LWAGE

1. You have data on years of work experience, EXPER, its square, EXPER2, years of education, EDUC, and the log of hourly wages, LWAGE 1. You have data on years of work experience, EXPER, its square, EXPER, years of education, EDUC, and the log of hourly wages, LWAGE You estimate the following regressions: (1) LWAGE =.00 + 0.05*EDUC +

More information

Testing Random Effects in Two-Way Spatial Panel Data Models

Testing Random Effects in Two-Way Spatial Panel Data Models Testing Random Effects in Two-Way Spatial Panel Data Models Nicolas Debarsy May 27, 2010 Abstract This paper proposes an alternative testing procedure to the Hausman test statistic to help the applied

More information

INFERENCE APPROACHES FOR INSTRUMENTAL VARIABLE QUANTILE REGRESSION. 1. Introduction

INFERENCE APPROACHES FOR INSTRUMENTAL VARIABLE QUANTILE REGRESSION. 1. Introduction INFERENCE APPROACHES FOR INSTRUMENTAL VARIABLE QUANTILE REGRESSION VICTOR CHERNOZHUKOV CHRISTIAN HANSEN MICHAEL JANSSON Abstract. We consider asymptotic and finite-sample confidence bounds in instrumental

More information

1. The Multivariate Classical Linear Regression Model

1. The Multivariate Classical Linear Regression Model Business School, Brunel University MSc. EC550/5509 Modelling Financial Decisions and Markets/Introduction to Quantitative Methods Prof. Menelaos Karanasos (Room SS69, Tel. 08956584) Lecture Notes 5. The

More information

PhD/MA Econometrics Examination January 2012 PART A

PhD/MA Econometrics Examination January 2012 PART A PhD/MA Econometrics Examination January 2012 PART A ANSWER ANY TWO QUESTIONS IN THIS SECTION NOTE: (1) The indicator function has the properties: (2) Question 1 Let, [defined as if using the indicator

More information

Fixed Effects Models for Panel Data. December 1, 2014

Fixed Effects Models for Panel Data. December 1, 2014 Fixed Effects Models for Panel Data December 1, 2014 Notation Use the same setup as before, with the linear model Y it = X it β + c i + ɛ it (1) where X it is a 1 K + 1 vector of independent variables.

More information

Økonomisk Kandidateksamen 2004 (I) Econometrics 2. Rettevejledning

Økonomisk Kandidateksamen 2004 (I) Econometrics 2. Rettevejledning Økonomisk Kandidateksamen 2004 (I) Econometrics 2 Rettevejledning This is a closed-book exam (uden hjælpemidler). Answer all questions! The group of questions 1 to 4 have equal weight. Within each group,

More information

ECON 4160, Spring term 2015 Lecture 7

ECON 4160, Spring term 2015 Lecture 7 ECON 4160, Spring term 2015 Lecture 7 Identification and estimation of SEMs (Part 1) Ragnar Nymoen Department of Economics 8 Oct 2015 1 / 55 HN Ch 15 References to Davidson and MacKinnon, Ch 8.1-8.5 Ch

More information

What s New in Econometrics? Lecture 14 Quantile Methods

What s New in Econometrics? Lecture 14 Quantile Methods What s New in Econometrics? Lecture 14 Quantile Methods Jeff Wooldridge NBER Summer Institute, 2007 1. Reminders About Means, Medians, and Quantiles 2. Some Useful Asymptotic Results 3. Quantile Regression

More information

1. The OLS Estimator. 1.1 Population model and notation

1. The OLS Estimator. 1.1 Population model and notation 1. The OLS Estimator OLS stands for Ordinary Least Squares. There are 6 assumptions ordinarily made, and the method of fitting a line through data is by least-squares. OLS is a common estimation methodology

More information

Beyond the Target Customer: Social Effects of CRM Campaigns

Beyond the Target Customer: Social Effects of CRM Campaigns Beyond the Target Customer: Social Effects of CRM Campaigns Eva Ascarza, Peter Ebbes, Oded Netzer, Matthew Danielson Link to article: http://journals.ama.org/doi/abs/10.1509/jmr.15.0442 WEB APPENDICES

More information

Linear Regression. Junhui Qian. October 27, 2014

Linear Regression. Junhui Qian. October 27, 2014 Linear Regression Junhui Qian October 27, 2014 Outline The Model Estimation Ordinary Least Square Method of Moments Maximum Likelihood Estimation Properties of OLS Estimator Unbiasedness Consistency Efficiency

More information

1 Outline. 1. Motivation. 2. SUR model. 3. Simultaneous equations. 4. Estimation

1 Outline. 1. Motivation. 2. SUR model. 3. Simultaneous equations. 4. Estimation 1 Outline. 1. Motivation 2. SUR model 3. Simultaneous equations 4. Estimation 2 Motivation. In this chapter, we will study simultaneous systems of econometric equations. Systems of simultaneous equations

More information

A Note on Demand Estimation with Supply Information. in Non-Linear Models

A Note on Demand Estimation with Supply Information. in Non-Linear Models A Note on Demand Estimation with Supply Information in Non-Linear Models Tongil TI Kim Emory University J. Miguel Villas-Boas University of California, Berkeley May, 2018 Keywords: demand estimation, limited

More information

Birkbeck Working Papers in Economics & Finance

Birkbeck Working Papers in Economics & Finance ISSN 1745-8587 Birkbeck Working Papers in Economics & Finance Department of Economics, Mathematics and Statistics BWPEF 1809 A Note on Specification Testing in Some Structural Regression Models Walter

More information

Econometrics of Panel Data

Econometrics of Panel Data Econometrics of Panel Data Jakub Mućk Meeting # 1 Jakub Mućk Econometrics of Panel Data Meeting # 1 1 / 31 Outline 1 Course outline 2 Panel data Advantages of Panel Data Limitations of Panel Data 3 Pooled

More information

Linear Panel Data Models

Linear Panel Data Models Linear Panel Data Models Michael R. Roberts Department of Finance The Wharton School University of Pennsylvania October 5, 2009 Michael R. Roberts Linear Panel Data Models 1/56 Example First Difference

More information

Quantile Regression for Dynamic Panel Data

Quantile Regression for Dynamic Panel Data Quantile Regression for Dynamic Panel Data Antonio Galvao 1 1 Department of Economics University of Illinois NASM Econometric Society 2008 June 22nd 2008 Panel Data Panel data allows the possibility of

More information

Estimation of Dynamic Nonlinear Random E ects Models with Unbalanced Panels.

Estimation of Dynamic Nonlinear Random E ects Models with Unbalanced Panels. Estimation of Dynamic Nonlinear Random E ects Models with Unbalanced Panels. Pedro Albarran y Raquel Carrasco z Jesus M. Carro x June 2014 Preliminary and Incomplete Abstract This paper presents and evaluates

More information

Vector Auto-Regressive Models

Vector Auto-Regressive Models Vector Auto-Regressive Models Laurent Ferrara 1 1 University of Paris Nanterre M2 Oct. 2018 Overview of the presentation 1. Vector Auto-Regressions Definition Estimation Testing 2. Impulse responses functions

More information

VAR Models and Applications

VAR Models and Applications VAR Models and Applications Laurent Ferrara 1 1 University of Paris West M2 EIPMC Oct. 2016 Overview of the presentation 1. Vector Auto-Regressions Definition Estimation Testing 2. Impulse responses functions

More information

Some Non-Parametric Identification Results using Timing and Information Set Assumptions

Some Non-Parametric Identification Results using Timing and Information Set Assumptions Some Non-Parametric Identification Results using Timing and Information Set Assumptions Daniel Ackerberg University of Michigan Jinyong Hahn UCLA December 31, 2015 PLEASE DO NOT CIRCULATE WITHOUT PERMISSION

More information

Problem Set #6: OLS. Economics 835: Econometrics. Fall 2012

Problem Set #6: OLS. Economics 835: Econometrics. Fall 2012 Problem Set #6: OLS Economics 835: Econometrics Fall 202 A preliminary result Suppose we have a random sample of size n on the scalar random variables (x, y) with finite means, variances, and covariance.

More information

Applied Microeconometrics (L5): Panel Data-Basics

Applied Microeconometrics (L5): Panel Data-Basics Applied Microeconometrics (L5): Panel Data-Basics Nicholas Giannakopoulos University of Patras Department of Economics ngias@upatras.gr November 10, 2015 Nicholas Giannakopoulos (UPatras) MSc Applied Economics

More information

Panel Threshold Regression Models with Endogenous Threshold Variables

Panel Threshold Regression Models with Endogenous Threshold Variables Panel Threshold Regression Models with Endogenous Threshold Variables Chien-Ho Wang National Taipei University Eric S. Lin National Tsing Hua University This Version: June 29, 2010 Abstract This paper

More information

HETEROSKEDASTICITY, TEMPORAL AND SPATIAL CORRELATION MATTER

HETEROSKEDASTICITY, TEMPORAL AND SPATIAL CORRELATION MATTER ACTA UNIVERSITATIS AGRICULTURAE ET SILVICULTURAE MENDELIANAE BRUNENSIS Volume LXI 239 Number 7, 2013 http://dx.doi.org/10.11118/actaun201361072151 HETEROSKEDASTICITY, TEMPORAL AND SPATIAL CORRELATION MATTER

More information

GMM estimation of spatial panels

GMM estimation of spatial panels MRA Munich ersonal ReEc Archive GMM estimation of spatial panels Francesco Moscone and Elisa Tosetti Brunel University 7. April 009 Online at http://mpra.ub.uni-muenchen.de/637/ MRA aper No. 637, posted

More information

-redprob- A Stata program for the Heckman estimator of the random effects dynamic probit model

-redprob- A Stata program for the Heckman estimator of the random effects dynamic probit model -redprob- A Stata program for the Heckman estimator of the random effects dynamic probit model Mark B. Stewart University of Warwick January 2006 1 The model The latent equation for the random effects

More information

Panel Data Exercises Manuel Arellano. Using panel data, a researcher considers the estimation of the following system:

Panel Data Exercises Manuel Arellano. Using panel data, a researcher considers the estimation of the following system: Panel Data Exercises Manuel Arellano Exercise 1 Using panel data, a researcher considers the estimation of the following system: y 1t = α 1 + βx 1t + v 1t. (t =1,..., T ) y Nt = α N + βx Nt + v Nt where

More information

Ordinary Least Squares Regression

Ordinary Least Squares Regression Ordinary Least Squares Regression Goals for this unit More on notation and terminology OLS scalar versus matrix derivation Some Preliminaries In this class we will be learning to analyze Cross Section

More information